Input Methods


I recently learned Shuangpin (double pinyin). While looking things up I was surprised how little Traditional Chinese material there is, so I put this together as notes—and to share with anyone who needs it. This post may suit one of these three kinds of readers:

  1. People who use Zhuyin and want to learn or understand Traditional Chinese pinyin input.
  2. People who already use pinyin and want to switch to Shuangpin—an advanced form of pinyin input.
  3. People who type English fast and Chinese slowly. (I’m guessing Chinese isn’t their first language?)

Why learn pinyin input? Aside from my own reasons, the biggest general perk is that typing Chinese also trains English typing, and typing English trains Chinese. Compared with people who only use Zhuyin or only type English, switching to the other usually feels sticky.

I also stumbled on an extra perk: if you have foreign friends who’ve studied Chinese (for some reason, whenever I hear the word “friend,” I picture Mayor Han giving a speech), and they want to write Chinese to you but hit a character they can’t write, they’ll spell it in pinyin—and you’ll be surprised you can read it. In chat, when they can’t read a Chinese character, you send them the romanization and somehow they get it too.

Pinyin input has two downsides, though:

  1. Too many keystrokes—so a few days ago I switched to Shuangpin. It compresses pinyin into one consonant letter plus one vowel letter; every Chinese character is two keystrokes, fewer than Zhuyin or Boshiamy.
  2. The other classic downside is many duplicate candidates (homophones), so you have to pick characters. Setting shape-based IMEs aside, any sound-based method has this problem. With stronger machine learning, auto-selection is accurate enough that it’s no longer a big deal.

There are many Shuangpin schemes; the popular ones include Xiaohe, Sogou, Microsoft, and Jiajia. Sogou and Microsoft are very similar; the others mean relearning from scratch, so pick one early. I hear Xiaohe is the most popular in China; my old Apple machine at home didn’t have it, so I went with Jiajia.


How do you teach yourself?

It’s simple on paper. I made a Zhuyin–pinyin–Shuangpin chart; print it out to look things up at first, or put small stickers on the matching keys, then type the sample characters a few times—that counts as a first “I’ve learned it.” Actually being able to use it is still a bit further.

Download the chart

Shuangpin input method

Then, like training muscle or any other skill, results track weekly volume most (10 hours vs 1 hour). Second most important: don’t practice too rarely—“3 hours once a week” beats worse than “1 hour on Mon, Wed, and Fri.”

For practice material, grab any article. I estimate after 10–20 hours you can barely start chatting—though by then the other person may get impatient with your glacial replies. Stick with it instead of switching back to Zhuyin, and in a few weeks you’ll slowly feel the perks above.

Those hour estimates are only a reference. I was a Boshiamy user, then practiced full pinyin, and only recently switched to Shuangpin; this post was typed after a dozen-plus hours of that switch. Zhuyin users can go straight to Shuangpin—it may take a bit longer, and I suspect you miss that extra perk above. If you want that perk, or want to understand Mainland pinyin rules, practice traditional pinyin first, then Shuangpin.


Early this year I saw an article titled:

“‘You Don’t Necessarily Need to Code’ — Bill Gates: Future Companies Need Talent With These 3 Backgrounds”

It made me nervous for a second—has AI already pushed into programming territory? The piece itself only argued that other knowledge matters too, but it made me wonder how likely programmers are to be automated away, or how fast. AI is already showing up in coding environments, after all.

Practicing Shuangpin gave me another thought. Look at a modern keyboard: top left, QWERTY… That layout dates to early mechanical typewriters designed to reduce jams—1874-era tech. We’re in the 21st century. Information tech has come this far; AlphaGo left human champions with no chance. Why do humans still have to talk “efficiently” to machines through something as primitive as a keyboard?

starbucks

Tech giants have tried hard to ditch the outdated keyboard—Apple’s mouse, touchscreens, Kinect gesture recognition, Google’s AI voice input, Elon Musk’s brain–computer interfaces… Capital has thrown astonishing money at it, yet at work you still need a keyboard. You can’t fully rely on Siri and toss the keyboard in the trash.

So will programming jobs be replaced? Looking at that same old keyboard sitting between humans and machines, I comfort myself: “Probably not that soon.”